Add ranges to identifiers and code blocks in the ASTs #146
Conversation
baronfel
force-pushed
the
add-ranges-to-AST
branch
from
541019b to
cdf9e34
Compare
| @@ -47,82 +47,82 @@ type ArgParser() = | |||
| sbuf.ToString() | |||
|
|
|||
|
|
|||
| static member ParsePartial(cursor,argv,argSpecs:seq<ArgInfo>,?other,?usageText) = | |||
There was a problem hiding this comment.
These changes were all to address some signature file/implementation argument name mismatch warnings.
| @@ -6,8 +6,8 @@ open System.Collections.Generic | |||
| open System.Globalization | |||
| open FSharp.Text.Lexing | |||
|
|
|||
| type Ident = string | |||
| type Code = string * Position | |||
| type Ident = string * Range | |||
There was a problem hiding this comment.
the primary change is to add a Range to the various identifier/text structures, and from there flow the effects of that change up to things like error reporting.
|
|
||
| type Rule = (Ident * Ident list * Clause list) | ||
| type Macro = Ident * Regexp | ||
| type Clause = { Matcher: Regexp; Code: Code } |
There was a problem hiding this comment.
adding types around these structures instead of tuples helps readability
| @@ -26,9 +26,12 @@ spec: | |||
| { TopCode=$1;Macros=$2;Rules=$4;BottomCode=$5 } | |||
| } | |||
|
|
|||
| ident: | |||
There was a problem hiding this comment.
the parser for fslex files itself had to be updated to adjust to the new AST shapes, supplying ranges where the types required it, or returning a record instead of a bare tuple.
| @@ -15,30 +15,30 @@ open FSharp.Text.Lexing | |||
| let (|KeyValue|) (kvp:KeyValuePair<_,_>) = kvp.Key,kvp.Value | |||
|
|
|||
|
|
|||
| type Identifier = string | |||
| type Code = string * Position | |||
| type Identifier = string * Range | |||
There was a problem hiding this comment.
similarly to fslex, fsyacc's own AST needed to be augmented with ranges for primitive productions like identifiers, code blocks, terminals, and nonterminals. This likewise flowed up through the remainder of the AST shapes.
|
|
||
| let checkNonTerminal (name, range) = | ||
| if name <> "error" && not (nonTerminalNamesSet.Contains name) then | ||
| failwithf "%s(%d,%d): NonTerminal '%s' has no productions" range.startPos.FileName range.startPos.Line range.startPos.Column name |
There was a problem hiding this comment.
this is an example of the nicer errors we can get from these structures. before this is merged I definitely want to convert the entire .Core api surface area to return Results containing all collected errors, though. This won't be usable inside an IDE if it throws all over the place.
|
something's different in the parsing of the ml-compatibility tests (the most recent commit failures should show this), but I'm not sure how concerned I should be. It might just be equality suffering due to the inclusion of the ranges on the identifiers? |
This PR changes the definitions of the types in the ASTs for FsLex and FsYacc. The intent would be to contain enough information to annotate identifiers in editor tooling to power experiences like code completion of identifiers, go-to-declaration, find usages, etc.
I've built this and started brief integrations into FsAutocomplete and initial results are promising! This would hopefully open to door to experiences like typechecking the TopCode/BottomCode, the individual rule productions, etc.